"St Lawrence of Google"

mcho writes "The Economist has a story about Google's co-founder, Larry Page, who " always wanted to change the world". The article attempts to make an arguement about the company's true intentions, amid all the rumors about potential Google products. "Google is already working on a massive and global computing grid. Eventually, says Mr Saffo, 'they're trying to build the machine that will pass the Turing test' -- in other words, an artificial intelligence that can pass as a human in written conversations. Wisely or not, Google wants to be a new sort of deus ex machina.""

Comments

Clutter

[Douglas Simmons]

A clean and uncluttered interface was the key to Google's search success as well as being the key supplement to their ad brokering business. I just hope "cluttering" up their business model won't have the opposite effect.

Don't mess with the missionary man

[chriss]

Forget about the AI rumor. It's just a rumor, the last sentence of TFA, unrelated to the rest.

More interesting is the following quote:

One visitor to the company's "Googleplex" in Silicon Valley "felt as if I were in the company of missionaries". A consequence of the theory that Google is aiming to run the world could be that "Google may be less liked in the industry than Microsoft inside 12 months," says Pip Coburn, a technology analyst. Bloggers have started accusing Google of hubris and arrogance.

This somehow reminds me of Apple in the 90s. They were on a crusade. They had found the holy grail. They could not fail. They would bring their vision to the world.

They could fail. And they failed. It didn't destroy them, but put their feet back to the ground. Where they belong. Today they make great products while listening to their users needs. They have learned that even though they may be on a mission, missionaries usually do not change the world. Hard workers and creative people do, as long as they stay connected to reality.

Bill Gates from Triumph of the nerds:

Success is a menace -- it fools smart people into thinking they can't lose.

Chriss

--
memomo.net - brush up your German, French, Spanish or Italian - online and free

Re:Don't mess with the missionary man

[JWW]

Dang, I just have to respond to this, and lose my mod points in this discussion...

Bill Gates from Triumph of the nerds:

        Success is a menace -- it fools smart people into thinking they can't lose.


That is absolutely the perfect quote to describe why Microsoft is the unbelivably paranoid company that it is. Bill always thinks Microsoft might lose and does any and everything (legal or not) to make sure that they don't.

This is what concerns me

[TallMatthew]

Paul Saffo at Silicon Valley's Institute for the Future says that "Google is a religion posing as a company."

I'm not exactly sure where a guy from a place called the "Institute for the Future" gets the nuts to call any organization pious, but he raises a point.

It's impossible to create a cathedral from a bazaar and still have it be a bazarr. You cannot suck all the resources out of the community and then declare yourself the community, which may or may not be Google's intent, but it certainly is starting to feel that way. They are chasing after every talented person around and positioning themselves in every market. Doing it better in some cases, not so much in others.

It's arguable, but innovation and competition seem to go hand in hand. We seem to produce better results when talent is spread around and several companies are chasing results, rather than one company gobbling everything up and amassing a vast fortune. I don't think Google is evil, but they may be too powerful for their own good. These massive projects they're taking on could have long-lasting effects in our community; I'd rather they were created in a consortium than in a star chamber.

Turing Test is dumb

[lawpoop]

I'm no computer scientist or whatever, but I think the Turing test is dumb.

My sig line says it all. Quoting Pablo Picasso: "Computers are useless. They can only give you answers" (Translated from Portugese, I guess).

So what if a computer can hold up it end of a conversation? What would be useful is if we could get a computer that wonders. Why is the sky blue? Why does it get dark at night? Where did I come from? How can I prove to someone else that I am conscious? How I do know that I'm conscious?

We have pretty decent tools of getting answers. What would really be a jump in human development is if we could get a machine to ask useful quesions.

Scary

[mtrupe]

People who view themselves as do-gooders and take so much pride in their righteousness scare me. If history has taught us anything, generally people who think they have a monopoly on good tend to end up doing bad.

Buy long term puts

[Budenny]

Whenever you see stories like this, and among other things if they start building new buildings, buy executive jets, if in Europe, CEOs get enobled, which is a particularly horrifying portent for shareholders, but if in the US start being treated as visionaries, then, buy long term puts. Especially when the brokerage community is telling you to buy buy buy.

Now, the great mistake in these matters is buying your puts too early, and I admit to thinking the time had come at 400. However, how anyone can lose in long term puts at this point defies belief. Is 500 possible? Probably. But I confidently expect to see 50 before we see 1,000. Friends, what we are seeing now is not part of the history of Internet or computing. It is a chapter in the history of hysteria.

Caution: this is not investment advice, and I am completely unqualified to give any. These are opinions offered to stimulate thought and discussion and of educational value only. If that!

More apt than originally intedned

[coolGuyZak]

FTA (emphasis added):

One visitor to the company's Googleplex in Silicon Valley felt as if I were in the company of missionaries. [...] Paul Saffo at Silicon Valley's Institute for the Future says that Google is a religion posing as a company. [...] If Google is a religion, what is its God? It would have to be The Algorithm. Faith in the possibility of an omniscient and omnipotent algorithm appears to be what Messrs Page and Brin have in common.

Google is not a software or hardware company

[scolby]

Google is an advertising company marketing themselves through cool free software. They've found a niche, and it's a good one. The idea that they're going to start producing operating systems or desktops is asinine...although I'm sure they will continue to donate to innovative initiatives like MIT's $200 computer, as doing so is also an excellent form of advertising and allows them actively to "not be evil."

To the naysayers... it's inevitable

[Kevin143]

So Google's big project is scanning every single book and indexing them online. It's a great idea. Why just search the internet when you can also be searching every work of literature? It's an obvious advance for Google, improving the search engine in a small but obvious way that makes a big difference as far as real usability.

Here's the thing: indexing books online is an incidental benefit. Google's real goal is to create a working, statistical AI. They've been hiring top-of-their-field AI researchers for a while. Last summer, Google won a competition for machine translation. They translated from Arabic to English and vice-versa better than all of their competitors. They did this using a statistical approach -- just feed the computer thousands and thousands of already translated documents, and eventually the machine can start making inferences based on probability. Given enough data, it works.

The same idea can be applied in the generic case. Wouldn't being able to ask an AI any question and receive a correct answer revolutionize society? And, the sum total of world literature is probably enough data to do so. They could call it AskG. He would know everything. And, the way they could roll it out, is by launching, and simultaneously updating wikipedia. It's well known that Wikipedia is riddled with small errors. Hell, the other day I inserted a gibberish statistic in an article about a city, and it's still there. Imagine if Google AI launches, and then announces that it has fixed Wikipedia. If Google AI made 50,000 edits it would overwhelm Wikipedia's normal editors, but whichever edits were checked by humans would certainly be confirmed as correct.

And, a new age of humanity would be ushered in. It would we a new Library of Alexandria. We would end the Age of Information and enter the Age of Knowledge. The singularity has already begun, but no one has realized it -- the singularity began the day Google went live.

Would AskG immediately fix quantum theory? Given all the data about science published by researchers, could G form new conclusions that humanity's best and brightest haven't? Could G solve the logistical challenge of solving world poverty?

There'd be one question left unanswered, of course, the classic "Can entropy be reversed?." What would be really scary would be if G had an immediate answer.

See the best sci-fi short story ever written, Asimov's The Last Question, or a simple find and replace hack of that story, The Last Query.

Re:To the naysayers... it's inevitable

[Anonymous Coward]

The Logical Conclusion

When earth's last thesis is copied
From the theses that went before,
When idea from fact has departed
And bare-boned factlets shall bore,
When all joy shall have fled from study
And scholarship reign supreme;
When truth shall "baaa" on the hill crests
And no one shall dare to dream;

When all the good poems have been buried
With comment annoted in full
And art shall bow down in homage
To scholarship's zinc-plated bull,
When there shall be nothing to research
But the notes of annoted notes,
And Baalam's ass shall inquire
The price of imported oats;

Then no one shall tell him the answer
For each shall know the one fact
That lies in the special ass-ignment
From which he is making his tract.
So the ass shall sigh uninstructed
While each in his separate book
Shall grind for the love of grinding
And only the devil shall look.

-Ezra Pound

Re:To the naysayers... it's inevitable

[spif]

This is a tempting vision, but the problem is that while Google has certainly built a nice big database and a lot of useful algorithms for interpreting search queries, they still need a lot more components to build something like what you suggest. And building these components is, to say the least, not trivial. I'm pretty sure they are thinking about at least some of these, probably all of them. That doesn't mean they're on the verge of making it happen.

A few of the components are, in no particular order:

A better interface
Limiting search queries to text strings isn't going to cut it. They need an interface that allows more complex input, and dynamic interaction with the results. They also need algorithms to handle searches based on that input. Complex input would include video, audio and other media types, individually and in combination. They would need the ability to form complex connections between different types of information. You should be able to show Google a scene from a movie and have it answer questions about that movie and anything related to it, within the context of the movie. Unless of course you change the context, whether explicitly or subtly. Basically you should be able to have a conversation with it just like you would talk to a trusted friend or colleague - one who just happens to have perfect recall and a huge store of knowledge.

Better identification and categorization (grouping) of results
They have this to some extent, of course. But it needs be more sophisticated by many orders of magnitude. One approach which I believe they are already using is to have trusted people provide feedback on automated categorization, kind of like meta-moderating on Slashdot only the original moderation is done by software. But this is barely beginning. One obstacle to this is that many sources of information on the web are actively mislabeled or "over labeled" by their creators, maintainers and/or third parties (a.k.a. Search Engine Optimization). The opposite is also true of many other data sources: they aren't properly labeled or categorized and so they get overlooked. Google should be able to know whether a piece of information is relevant as easily as would an (open minded) expert who looked at it. And it should be able to tell you things like "this piece of information is related to what you're asking about, and it's sort of interesting, but it might be a little crackpot because..." and so forth.

Digital limits
At any point in time there are limits to what information can be acquired, stored and accessed in digital form. Assuming Google is limiting itself to dealing with information in digital form, these limits will apply. Some classes of information are not available to Google for reasons of personal privacy or proprietary business interest. Other information is simply beyond the ability of current technology to capture in digital form, like unexpressed or uncaptured human thought and emotion. This includes nuance. It's not a minor thing to lack nuance, either, because something like 90% or more of human communication is estimated to consist of things like vocal tone and facial expression.

Anyhow, I'm no expert, but I would say your vision is far from reality. Like I said, I suspect Google is moving towards something like what you suggest, but I don't think it is forthcoming in the near future. More likely is that progress in these areas will lead to better techniques for capturing opinion, nuance and subtlety that is related to "hard" information. These are things that will still require vast amounts of human input, networks of trust, etc. But Google will hopefully help automate a lot of the "legwork" in that process. If not, someone else probably will, because the result will be an incredibly valuable (and therefore lucrative) tool for business and personal advantage.

Re:To the naysayers... it's inevitable

[vinlud]

I don't think Googles future lies in AI. They make money by building a huge worldwide database of online user profiles which can be used for selling advertisements. How to gather all this necessary information? By providing free useful tools everybody wants, like searching the web or books, maps, email, etcetera. By the way I heard Google will launch a free web statistics tool quite soon, which would be a logical step in their goal to aquire as much information for profiles as possible.

Two concerns about Google

[Anonymous Coward]

This is a funny article. It has no new content. Disappointing for The Economist which usually has good stuff. Two other things:

1. One thing that bothers me about Google is the fact that they are collecting all this data about our searches. Like Scott McNeally said some time back, I do realize that "there is no such thing as Privacy -- get over it". But I still don't like it. The strange thing about Google is that I believe that they ARE competent enough to misuse all this data (Microsoft may be evil but they are also incompetent). And they are more and more commercial (meaning, less and less ethical). So I have stopped using Google exclusively for my searches and have started spreading them over MSN and Ask Jeeves (which is Yahoo Search with a cleaner interface).

2. As I use other search engines more and more, I have noticed that they are beginning to have results *almost* as good as Google. Which brings me to my next point: Google has not done anything breathtakingly good in a long time -- I know, I know -- Google Map, Gmail etc are good -- but not *breakthrough* products like Google Search was when I first used it in the summer of 1999. The latest example is the Google Pack, a very underwhelming product. And the way Google Videostore was introduced last week, with all the store-is-up-now-store-is-down-now problems, was a clearly hamhanded, premature way to introduce a new product. This should be of concern to people who own Google stock. I think there is a problem here. Are they losing their mojo?

Re:AI - Where will it come from?

[ltbarcly]

This is just platitudes and idiocy.

The natural state of matter is not consciousness. If it were virtually everything would be intelligent. As it is only a few animals seem to possess intelligence on one planet.

We know that consciousness in Man is the result of billions of years of competition among trillions upon trillions of organisms which are our ancestors.

The idea that a single entity, designed, but not designed to be conscious will eventually become intelligent is the result of too much bad science fiction. Trillions of organisms evolving for billions of years to produce even slightly intelligent animals vs. a single network with much less than a billion nodes and no evolutionary forces at work whatsoever.

AI will be developed when we unravel the secrets to intelligence or when we produce enormously fast computer simulated evolution, but it will not come about as a side effect of people surfing porn.

Re:T1,2,3

[pipingguy]

Marx did not foresee computers and data becoming so valuable.

Neither did capitalism.

Re:Turing test answers that one

[lawpoop]

I guess I have another side to my point.

Well, if we come up with a device that passes the turing test, that doesn't necessarily tell us anything about the human mind. It could be that there is only one implementation of an intelligence or consciousness or whatever, or the machine we invent could be completely different from the human mind, yet achieve the same results.

If you don't care about how the human mind specifically works, that's fine. But my point is that if a machine passes the Turing test, that doesn't *necessarily* mean anything about how the human mind works.